RAAIS 2017 - Simon Knowles, Co-Founder and CTO of Graphcore
Channel:The Research and Applied AI Summit - RAAIS · Length:35:44 · Published:2017-07-13 · Views:6896
Summary
Simon Knowles, co-founder and CTO of Graphcore, talks about "intelligence processors", chips designed specifically for machine learning. The talk was given at the RAAIS 2017 summit and covers Graphcore's architecture.
Why HN recommends it
HN points to it as a technical talk from a chip designer on why AI hardware should put memory close to compute and use async data transfers.
Here's a nice technical talk from Graphcore's CTO about that — HN, 2017
Machine learning & deep learning · Computer architecture & CPUs
Subtitles (excerpt)
up next I'm super happy to welcome Simon knows who's the CTO and co-founder of Groff core which is uh inventing a new processor for the machine learning age Simon's been in the business for a very long time founded two companies I Sarah and Ellen 14 resold for over a billion so you knows what to do here so over over to you Simon do I don't go away oh thank you so graph core is a relatively young company we announced our existence and our first round of funding last September we have about 50 people today pretty much all engineers we've actually been working on this hardware I'll say a little bit about for about three years om released specifications yet I'm not going to do a sort of full show-and-tell but I'll present some outline and some discussion of the sorts of things that are guiding our mission and in fact we we will make another funding related announced very shortly now we are growing like a weed if anyone fancies a career in chips for machine learning you know where to come at the moment we are most famous for producing some eye candy that pops out of our tools this is a resonate layer-by-layer evolution you can see some of the stacks of identical layers of convolutional nets go around these loops here we gain no insight from this this is just I candy
On Hacker News
I went to a talk from the CTO of Graphcore ( [LINK] ) on Monday. They are designing chips targeted at machine learning. As I understood it, their architecture comprises - lots of "tiles" - small processing cores with collocated memory (essentially DSPs) - very high bandwidth (90TB/s!) switching fabric to move data between tiles - "Bulk Synchronous Parallel" operation, meaning that the tiles do th… — maffydub, 2018-04
Indeed, and that's why there are a couple of startups working on new chips and why Google has the TPU. Here's a nice technical talk from Graphcore's CTO about that [VIDEO] — midko, 2017-11
Interesting approach, but very far off. I prefer GraphCore as a promising new lead in AI hardware. Moving computation close in with memory, and making data transfers async will speed up DL. [VIDEO] — visarga, 2017-07
Related videos
- RL Course by David Silver - Lecture 1: Introduction to Reinforcement Learning · Google DeepMind · 1:28:12
- Why Greatness Cannot Be Planned — Kenneth Stanley · Machine Learning Street Talk · 2:46:27
- High-Speed Robot Hand · Hizook · 3:00
- Neural Network Learns The Physics of Fluids and Smoke | Two Minute Papers #118 · Two Minute Papers · 5:26
- Attention Is All You Need · Yannic Kilcher · 27:06
- The spelled-out intro to neural networks and backpropagation: building micrograd · Andrej Karpathy · 2:25:52
- But what is a neural network? | Deep learning chapter 1 · 3Blue1Brown · 18:40
- 8-bit computer update · Ben Eater · 6:53
- Light Years Ahead | The 1969 Apollo Guidance Computer · TNMoC · 1:21:22
- “Hello, world” from scratch on a 6502 — Part 1 · Ben Eater · 27:24
- AlphaGo - The Movie | Full award-winning documentary · Google DeepMind · 1:30:28
- Transformers, the tech behind LLMs | Deep Learning Chapter 5 · 3Blue1Brown · 27:14
- Finding the Missing Memristor - R. Stanley Williams · The Qualcomm Institute · 47:52
- The Ultimate Game Boy Talk (33c3) · media.ccc.de · 1:01:06
- USENIX Security '18-Q: Why Do Keynote Speakers Keep Suggesting That Improving Security Is Possible? · USENIX · 51:22
- The Thirty Million Line Problem · Molly Rocket · 1:48:55